Risk Groups in Penile Cancer

Aim of the study

The aim of this study is to evaluate the accuracy of previously published risk groups systems for predicting inguinal nodal metastases in patients with penile carcinoma. Cases of invasive penile squamous cell carcinomas were stratified using the following systems: Solsona et al (J Urol 2001;165:1506), Hungerhuber et al (Urology 2006;68:621), and the proposed by the European Association of Urology (Eur Urol 2004;46:1), with low, intermediate, and high risk categories in each one of them. Metastatic rates and cancer-specific survival rates in our dataset were compared with previously reported results. Receiver-operator characteristic (ROC) analysis was carried out to compare accuracy in predicting final nodal status. We found that these risk groups systems may be useful for patients with low-grade superficial tumors and less accurate for evaluating patients with high-grade locally-advanced penile carcinomas. The results of this study may be useful for therapeutic planning of patients with penile squamous cell carcinomas.

Description of the repository

This repository contains the full statistical analysis of the dataset that was used for the article “Risk Groups Systems for Penile Cancer Management: A Study of 203 Patients with Invasive Squamous Cell Carcinoma”. The article has been accepted for publication in Urology. This repository also contains the following files:

  • The final (approved) PDF version of the article, as submitted for publication
  • The BibTeX file containing all the references cited in the article
  • The R script that was used for analyzing the dataset and write the article, as well as the R scripts containing the functions written for plotting figures and tables
  • The R Markdown file used for this report
  • The figures included in this repository in PNG format

Data were analyzed using R version 3.1.1 “Sock it to Me” (R Foundation for Statistical Computing, Vienna, Austria). Results were written using RMarkDown in RStudio version 0.98.1102 and the knitr package version 1.9 by Yihui Xie.

Building the dataset for analysis

First we loaded the full dataset including 333 patiens with invasive penile squamous cell carcinoma.

Data <- read.csv("Article/PenisSCC_333.csv")

The full dataset is available at http://dx.doi.org/10.6084/m9.figshare.1290997, which also contains the dataset’s codebook. From the 333 patients we selected only those with total/partial penectomy in Procedure, excluded cases with missing values in the variables of Grade, pT and cN, and excluded patients who were lost at follow up in Outcome.

Data <- subset(Data, Procedure == "Total penectomy" | Procedure == "Partial penectomy")
Data <- Data[complete.cases(Data$Grade), ]
Data <- Data[complete.cases(Data$pT), ]
Data <- Data[complete.cases(Data$cN), ]
Data <- subset(Data, Outcome != "Lost at follow-up")
Data <- droplevels(Data)

We then recoded 3 variables, creating new variables for cN positivity (cN_Positive), cancer-related death (DOD), and final nodal status (Final_Nodal). Positive cases for Final_Nodalwhere those with lymph node metastasis in the groin dissection, local relapse during follow-up, or unfavorable outcome, including alive with disease and death by cancer.

Data$cN_Positive <- ifelse(Data$cN == "cN0", c("No"), c("Yes"))
Data$cN_Positive <- as.factor(Data$cN_Positive)
Data$DOD <- ifelse(Data$Outcome == "Died of cancer", c("Yes"), c("No"))
Data$DOD <- as.factor(Data$DOD)
Data$Final_Nodal <- ifelse(Data$Mets == "Yes", c("Positive"), c("Negative"))
Data$Final_Nodal[Data$Local == "Yes"] <- "Positive"
Data$Final_Nodal[Data$Outcome == "Alive with disease"] <- "Positive"
Data$Final_Nodal[Data$Outcome == "Died of cancer"] <- "Positive"
Data$Final_Nodal <- as.factor(Data$Final_Nodal)

We finally created the risk groups for the Solsona et al, EUA, and Hungerhuber et al systems, using the criteria provided in the reports.

# SOLSONA risk groups
Data$Solsona[Data$Grade == "Grade 1" & Data$pT == "T1"] <- "Low risk"
Data$Solsona[Data$Grade == "Grade 2" & Data$pT == "T1"] <- "Intermediate risk"
Data$Solsona[Data$Grade == "Grade 3" & Data$pT == "T1"] <- "Intermediate risk"
Data$Solsona[Data$Grade == "Grade 1" & Data$pT == "T2"] <- "Intermediate risk"
Data$Solsona[Data$Grade == "Grade 1" & Data$pT == "T3"] <- "Intermediate risk"
Data$Solsona[Data$Grade == "Grade 2" & Data$pT == "T2"] <- "High risk"
Data$Solsona[Data$Grade == "Grade 3" & Data$pT == "T2"] <- "High risk"
Data$Solsona[Data$Grade == "Grade 2" & Data$pT == "T3"] <- "High risk"
Data$Solsona[Data$Grade == "Grade 3" & Data$pT == "T3"] <- "High risk"
Data$Solsona <- factor(Data$Solsona, levels = c("Low risk", "Intermediate risk", "High risk"), ordered = TRUE)
# EAU risk groups
Data$EAU[Data$Grade == "Grade 1" & Data$pT == "T1"] <- "Low risk"
Data$EAU[Data$Grade == "Grade 2" & Data$pT == "T1"] <- "Intermediate risk"
Data$EAU[Data$Grade == "Grade 3" & Data$pT == "T1"] <- "High risk"
Data$EAU[Data$pT == "T2"] <- "High risk"
Data$EAU[Data$pT == "T3"] <- "High risk"
Data$EAU <- factor(Data$EAU, levels = c("Low risk", "Intermediate risk", "High risk"), ordered = TRUE)
# HUNGERHUBER risk groups
Data$Hungerhuber[Data$Grade == "Grade 1" & Data$pT == "T1"] <- "Low risk"
Data$Hungerhuber[Data$Grade == "Grade 2" & Data$pT == "T1"] <- "Low risk"
Data$Hungerhuber[Data$Grade == "Grade 1" & Data$pT == "T2"] <- "Intermediate risk"
Data$Hungerhuber[Data$Grade == "Grade 1" & Data$pT == "T3"] <- "Intermediate risk"
Data$Hungerhuber[Data$Grade == "Grade 2" & Data$pT == "T2"] <- "Intermediate risk"
Data$Hungerhuber[Data$Grade == "Grade 2" & Data$pT == "T3"] <- "Intermediate risk"
Data$Hungerhuber[Data$Grade == "Grade 3"] <- "High risk"
Data$Hungerhuber <- factor(Data$Hungerhuber, levels = c("Low risk", "Intermediate risk", "High risk"), ordered = TRUE)

Data analysis was carried out on this dataset, using the simpleR package.

library(simpleR)

Methodology

Data analysis is divided in 4 sections, as it follows:

Descriptive Statistics. All the variables included in the dataset were analyzed using bar plots, histograms, box plots, and one-way tables. Factor variables were described using absolute and relative percentages. Numeric variables were described using mean, standard deviation, median, interquartile range, minimum and maximum value.

Inferential Statistics: Statistical tests (Fisher’s exact test for categorical variables, Kruskal-Wallis test for numerical variables) were used to evaluate the association between Final Nodal Status and Cancer-Related Death as predictors and all the variables included in the dataset. A 2-tailed P value was reported in all instances. Reported statistics included absolute and relative percentages for categorical variables; and mean, standard deviation, median, interquartile range, minimum and maximum value for numeric variables, by final nodal status and by cancer-related death.

Survival Analysis. For all variables in the dataset survival curves by Final Nodal Status and by Cancer-Related Death were built using the Kaplan-Meier method and compared using the Mantel-Cox (log-rank) test. Numerical variables were splitted in 2 levels using the median as the cutoff point. A 2-tailed P value was reported in all instances.

Receiver-Operator Characteristic (ROC) Curve Analysis. ROC curves for predicting Final Nodal Status and Cancer-Related Death were plotted for all risk groups systems. ROC curves were compared against the baseline using the Mann-Whitney U test. Reported statistics included the area under the ROC curve (AUC) with tis 95% confidence interval and the 2-tailed P value from the Mann-Whitney test.

Descriptive Statistics

Here it follows the description of all the variables included in the analyzed dataset.

Surgical procedure for primary treatment

Var <- Data$Procedure
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Partial penectomy 124 61
Total penectomy 79 39

Number of missing cases: 0 cases.

Histologic grade

Var <- Data$Grade
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Grade 1 52 26
Grade 2 69 34
Grade 3 82 40

Number of missing cases: 0 cases.

Histologic subtype

Var <- Data$Subtype
categorical.plot(Var, align = "h", left =8)

descriptive.categorical(Var)
No. Cases %
Adenosquamous 3 1.48
Basaloid 9 4.43
Mixed 21 10.34
Papillary 6 2.96
Sarcomatoid 2 0.99
Usual 132 65.02
Verrucous 17 8.37
Warty 13 6.40

Number of missing cases: 0 cases.

Tumor located in glans

Var <- Data$Glans
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 2 1
Yes 198 99

Number of missing cases: 3 cases.

Tumor located in coronal sulcus

Var <- Data$Sulcus
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 146 73
Yes 54 27

Number of missing cases: 3 cases.

Tumor located in Foreskin

Var <- Data$Foreskin
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 159 80
Yes 41 20

Number of missing cases: 3 cases.

Anatomical level of maximum tumor invasion

Var <- Data$Level
categorical.plot(Var, align = "h", left = 9)

descriptive.categorical(Var)
No. Cases %
Corpus cavernosum 122 60.1
Corpus spongiosum 65 32.0
Lamina propria 9 4.4
Preputial dartos 7 3.4

Number of missing cases: 0 cases.

Tumor thickness in mm

Var <- Data$Thickness
numerical.plot(Var, label = "Tumor Thickness, mm")

descriptive.numerical(Var)
Statistics Values
Mean 7.5
Standard Deviation 3.7
Median 7
Interquartile Range 5
Mininum 1
Maximum 20

Number of missing cases: 11 cases.

Tumor size in cm

Var <- Data$Size
numerical.plot(Var, label = "Tumor Size, cm")

descriptive.numerical(Var)
Statistics Values
Mean 4.7
Standard Deviation 2.5
Median 4
Interquartile Range 2.8
Mininum 1
Maximum 16

Number of missing cases: 91 cases.

Patient’s age in years

Var <- Data$Age
numerical.plot(Var, label = "Patient's Age, years")

descriptive.numerical(Var)
Statistics Values
Mean 53.9
Standard Deviation 14.1
Median 53
Interquartile Range 21
Mininum 24
Maximum 88

Number of missing cases: 0 cases.

Bilateral inguinal lymphadenectomy

Var <- Data$Lymphadenectomy
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 105 52
Yes 98 48

Number of missing cases: 0 cases.

Inguinal Lymph Node Metastasis

Var <- Data$Mets
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 152 75
Yes 51 25

Number of missing cases: 0 cases.

Tumor invasion of penile urethra

Var <- Data$Urethra
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 115 57
Yes 88 43

Number of missing cases: 0 cases.

Vascular invasion in primary tumor

Var <- Data$Vascular
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 141 72
Yes 55 28

Number of missing cases: 7 cases.

Perineural invasion in primary tumor

Var <- Data$Perineural
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 134 68
Yes 63 32

Number of missing cases: 6 cases.

Pathological T stage

Var <- Data$pT
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
T1 8 3.9
T2 107 52.7
T3 88 43.3

Number of missing cases: 0 cases.

Any tumor relapse (local, regional or systemic)

Var <- Data$Relapse
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 150 77
Yes 45 23

Number of missing cases: 8 cases.

Local tumor relapse

Var <- Data$Local
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 37 82
Yes 8 18

Number of missing cases: 158 cases.

Regional tumor relapse

Var <- Data$Regional
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 9 20
Yes 36 80

Number of missing cases: 158 cases.

Systemic tumor relapse

Var <- Data$Systemic
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 33 73
Yes 12 27

Number of missing cases: 158 cases.

Follow-up length in months

Var <- Data$FollowUp
numerical.plot(Var, label = "Follow-Up, Months")

descriptive.numerical(Var)
Statistics Values
Mean 108.1
Standard Deviation 94.7
Median 86.8
Interquartile Range 152.4
Mininum 0.8
Maximum 433.9

Number of missing cases: 0 cases.

Patient’s outcome

Var <- Data$Outcome
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Alive with disease 14 6.9
Alive without disease 89 43.8
Died of cancer 34 16.7
Died of other causes 66 32.5

Number of missing cases: 0 cases.

Clinical N stage

Var <- Data$cN
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
cN0 108 53.2
cN1 31 15.3
cN2 59 29.1
cN3 5 2.5

Number of missing cases: 0 cases.

Clinically positive inguinal lymph nodes

Var <- Data$cN_Positive
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
No 108 53
Yes 95 47

Number of missing cases: 0 cases.

Final nodal status

Var <- Data$Final_Nodal
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Negative 132 65
Positive 71 35

Number of missing cases: 0 cases.

Solsona et al risk groups

Var <- Data$Solsona
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Low risk 5 2.5
Intermediate risk 50 24.6
High risk 148 72.9

Number of missing cases: 0 cases.

European Association of Urology risk groups

Var <- Data$EAU
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Low risk 5 2.46
Intermediate risk 2 0.99
High risk 196 96.55

Number of missing cases: 0 cases.

Hungerhuber et al risk groups

Var <- Data$Hungerhuber
categorical.plot(Var)

descriptive.categorical(Var)
No. Cases %
Low risk 7 3.4
Intermediate risk 114 56.2
High risk 82 40.4

Number of missing cases: 0 cases.

Inferential Statistics: Final Nodal Status

Var2 <- Data$Final_Nodal

Final nodal status and surgical procedure

Var1 <- Data$Procedure
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
Partial penectomy 95 72 29 40.8
Total penectomy 37 28 42 59.2

Final nodal status and histologic grade

Var1 <- Data$Grade
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
Grade 1 48 36.4 4 5.6
Grade 2 46 34.8 23 32.4
Grade 3 38 28.8 44 62.0

Final nodal status and glans location

Var1 <- Data$Glans
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 2 1.6 0 0
Yes 127 98.4 71 100

Final nodal status and coronal sulcus location

Var1 <- Data$Sulcus
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 95 73.6 51 71.8
Yes 34 26.4 20 28.2

Final nodal status and foreskin location

Var1 <- Data$Glans
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 2 1.6 0 0
Yes 127 98.4 71 100

Final nodal status and anatomical level

Var1 <- Data$Level
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
Corpus cavernosum 68 51.5 54 76.1
Corpus spongiosum 50 37.9 15 21.1
Lamina propria 9 6.8 0 0.0
Preputial dartos 5 3.8 2 2.8

Final nodal status and tumor thickness

Var1 <- Data$Thickness
numerical.group.plot(Var1, Var2)

descriptive.numerical.group(Var1, Var2)
Negative Positive
Mean 6.8 8.7
Standard Deviation 3.3 4.1
Median 7.0 8.0
Interquartile Range 3.0 6.0
Minimum 1.0 2.0
Maximum 20.0 20.0

Final nodal status and tumor size

Var1 <- Data$Size
numerical.group.plot(Var1, Var2)

descriptive.numerical.group(Var1, Var2)
Negative Positive
Mean 4.4 5.2
Standard Deviation 2.4 2.7
Median 4.0 5.0
Interquartile Range 2.5 3.5
Minimum 1.5 1.0
Maximum 16.0 12.0

Final nodal status and patient’s age

Var1 <- Data$Age
numerical.group.plot(Var1, Var2)

descriptive.numerical.group(Var1, Var2)
Negative Positive
Mean 54.3 53.0
Standard Deviation 14.5 13.4
Median 53.0 53.0
Interquartile Range 22.0 19.0
Minimum 24.0 27.0
Maximum 88.0 82.0

Final nodal status and urethral invasion

Var1 <- Data$Urethra
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 84 63.6 31 43.7
Yes 48 36.4 40 56.3

Final nodal status and vascular invasion

Var1 <- Data$Vascular
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 104 81.2 37 54.4
Yes 24 18.8 31 45.6

Final nodal status and perineural invasion

Var1 <- Data$Perineural
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 109 85.2 25 36.2
Yes 19 14.8 44 63.8

Final nodal status and clinical N stage

Var1 <- Data$cN
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
cN0 92 69.7 16 22.5
cN1 13 9.8 18 25.4
cN2 25 18.9 34 47.9
cN3 2 1.5 3 4.2

Final nodal status and pathological T stage

Var1 <- Data$pT
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
T1 8 6.1 0 0.0
T2 76 57.6 31 43.7
T3 48 36.4 40 56.3

Final nodal status and (any) tumor relapse

Var1 <- Data$Relapse
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 126 95.5 24 38.1
Yes 6 4.5 39 61.9

Final nodal status and local relapse

Var1 <- Data$Local
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 6 100 31 79.5
Yes 0 0 8 20.5

Final nodal status and regional relapse

Var1 <- Data$Regional
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 0 0 9 23.1
Yes 6 100 30 76.9

Final nodal status and systemic relapse

Var1 <- Data$Systemic
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
No 6 100 27 69.2
Yes 0 0 12 30.8

Final nodal status and Solsona et al risk groups

Var1 <- Data$Solsona
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
Low risk 5 3.8 0 0.0
Intermediate risk 46 34.8 4 5.6
High risk 81 61.4 67 94.4

Final nodal status and EAU risk groups

Var1 <- Data$EAU
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
Low risk 5 3.8 0 0
Intermediate risk 2 1.5 0 0
High risk 125 94.7 71 100

Final nodal status and Hungerhuber et al risk groups

Var1 <- Data$Hungerhuber
categorical.group.plot(Var1, Var2)

descriptive.categorical.group(Var1, Var2)
Negative % Positive %
Low risk 7 5.3 0 0
Intermediate risk 87 65.9 27 38
High risk 38 28.8 44 62

Survival Analysis: Final Nodal Status

# Defining outcome variable
Status <- Data$Final_Nodal
# Creating dicotomic variables from numerical variables for plotting
Thickness_Median <- factor(ifelse(Data$Thickness > median(Data$Thickness, na.rm = TRUE), c("Above Median Thickness"), c("Below Median Thickness")))
Age_Median <- factor(ifelse(Data$Age > median(Data$Age, na.rm = TRUE), c("Above Median Age"), c("Below Median Age")))
# By surgical procedure
with(Data, survival.plot(Procedure, FollowUp, Status, title = "Final Nodal Status by Surgical Procedure"))

# By histologic grade
with(Data, survival.plot(Grade, FollowUp, Status, title = "Final Nodal Status by Histologic Grade", position = "bottomright"))

# By glans invasion
with(Data, survival.plot(Glans, FollowUp, Status, title = "Final Nodal Status by Glans Invasion"))

# By coronal sulcus invasion
with(Data, survival.plot(Sulcus, FollowUp, Status, title = "Final Nodal Status by Coronal Sulcus Invasion"))

# By foreskin invasion
with(Data, survival.plot(Foreskin, FollowUp, Status, title = "Final Nodal Status by Foreskin Invasion"))

# By anatomical level
with(Data, survival.plot(Level, FollowUp, Status, title = "Final Nodal Status by Anatomical Level"))

# By median tumor thickness
with(Data, survival.plot(Thickness_Median, FollowUp, Status, title = "Final Nodal Status by Median Tumor Thickness"))

# By median patient's age
with(Data, survival.plot(Age_Median, FollowUp, Status, title = "Final Nodal Status by Median Patient's Age"))

# By urethral invasion
with(Data, survival.plot(Urethra, FollowUp, Status, title = "Final Nodal Status by Urethral Invasion"))

# By vascular invasion
with(Data, survival.plot(Vascular, FollowUp, Status, title = "Final Nodal Status by Vascular Invasion"))

# By perineural invasion
with(Data, survival.plot(Perineural, FollowUp, Status, title = "Final Nodal Status by Perineural Invasion"))

# By pathological T stage
with(Data, survival.plot(pT, FollowUp, Status, title = "Final Nodal Status by Pathological T Stage"))

# By clinical N stage positivity
with(Data, survival.plot(cN_Positive, FollowUp, Status, title = "Final Nodal Status by Clinical N Stage Positivity"))

# By Solsona et al risk groups
with(Data, survival.plot(Solsona, FollowUp, Status, title = "Final Nodal Status by Solsona et al Risk Groups", position = "bottomright"))

# By EAU risk groups
with(Data, survival.plot(EAU, FollowUp, Status, title = "Final Nodal Status by EAU Risk Groups"))

# By Hungerhuber et al risk groups
with(Data, survival.plot(Hungerhuber, FollowUp, Status, title = "Final Nodal Status by Hungerhuber et al Risk Groups"))

ROC Curve Analysis: Final Nodal Status

library(pROC)
## Type 'citation("pROC")' for a citation.
## 
## Attaching package: 'pROC'
## 
## The following objects are masked from 'package:stats':
## 
##     cov, smooth, var
# Plots
par(cex = 1.25)
FN_ROC_Solsona <- with(Data, plot.roc(Final_Nodal, Solsona,  main = "Final Nodal Status", lty = 4, col = 4))
FN_ROC_EAU <- with(Data, lines.roc(Final_Nodal, EAU, lty = 2, col = 1))
FN_ROC_Hungerhuber <- with(Data, lines.roc(Final_Nodal, Hungerhuber, lty = 6, col = 2))
legend("bottomright", lty = c(4,2,6), col = c(4,1,2), bty = "n",
       legend = c("Solsona et al system", "EAU system", "Hungerhuber et al system"))

# Statistics
ROC_FN_Solsona <- round(with(Data, roc(Final_Nodal ~ Solsona, ci = TRUE)$ci), 2)
MW_FN_Solsona <- format(with(Data, wilcox.test(as.numeric(Solsona) ~ Final_Nodal)$p.value), digits = 2)
ROC_FN_EAU <- round(with(Data, roc(Final_Nodal ~ EAU, ci = TRUE)$ci), 2)
MW_FN_EAU <- format(with(Data, wilcox.test(as.numeric(EAU) ~ Final_Nodal)$p.value), digits = 2)
ROC_FN_Hungerhuber <- round(with(Data, roc(Final_Nodal ~ Hungerhuber, ci = TRUE)$ci), 2)
MW_FN_Hungerhuber <- format(with(Data, wilcox.test(as.numeric(Hungerhuber) ~ Final_Nodal)$p.value), digits = 2)
Risk Group System Area Under the ROC Curve 95% CI Mann-Whitney’s P value
Solsona et al system 0.67 0.62, 0.72 4.6e-07
EAU system 0.53 0.51, 0.55 0.049
Hungerhuber et al system 0.68 0.61, 0.74 2.1e-06